Face Recognition Security System Based on Convolutional Neural Networks
Face recognition is a complicated process due to the similarity of the overall structure of faces regardless of the slight differences between one face and another. Since face recognition systems are used in many fields such as security systems, it is necessary to find effective and low-complexity facial recognition methods. The proposed system presents a suitable face recognition method that uses an image histogram equalization that enhances the image, a wavelet transform that compresses the image, and uses using multi- neural networks that can handle all remaining changes such as rotation, scaling, and deformation. Finally, the decision unit is combined with outputs and a small set of images of each person in the database to deliver adequate classification accuracy. The designed system used an efficiently recognizes faces without being affected by distortion, rotation or angle change. The overall system is written and implemented as a software package using Python.